ffs_balloc.c revision 1.14.4.6 1 /* $NetBSD: ffs_balloc.c,v 1.14.4.6 1999/08/06 12:50:04 chs Exp $ */
2
3 /*
4 * Copyright (c) 1982, 1986, 1989, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)ffs_balloc.c 8.8 (Berkeley) 6/16/95
36 */
37
38 #if defined(_KERNEL) && !defined(_LKM)
39 #include "opt_quota.h"
40 #endif
41
42 #include <sys/param.h>
43 #include <sys/systm.h>
44 #include <sys/buf.h>
45 #include <sys/proc.h>
46 #include <sys/file.h>
47 #include <sys/vnode.h>
48 #include <sys/mount.h>
49
50 #include <vm/vm.h>
51 #include <uvm/uvm_extern.h>
52 #include <uvm/uvm.h>
53
54 #include <ufs/ufs/quota.h>
55 #include <ufs/ufs/ufsmount.h>
56 #include <ufs/ufs/inode.h>
57 #include <ufs/ufs/ufs_extern.h>
58 #include <ufs/ufs/ufs_bswap.h>
59
60 #include <ufs/ffs/fs.h>
61 #include <ufs/ffs/ffs_extern.h>
62
63 /*
64 * Balloc defines the structure of file system storage
65 * by allocating the physical blocks on a device given
66 * the inode and the logical block number in a file.
67 */
68
69 int
70 ffs_balloc(v)
71 void *v;
72 {
73 struct vop_balloc_args /* {
74 struct vnode *a_vp;
75 off_t a_offset;
76 off_t a_length;
77 struct ucred *a_cred;
78 int a_flags;
79 } */ *ap = v;
80
81 off_t off, len;
82 struct vnode *vp = ap->a_vp;
83 struct inode *ip = VTOI(vp);
84 struct fs *fs = ip->i_fs;
85 int error, delta, bshift, bsize;
86
87 bshift = fs->fs_bshift;
88 bsize = 1 << bshift;
89
90 off = ap->a_offset;
91 len = ap->a_length;
92
93 delta = off & (bsize - 1);
94 off -= delta;
95 len += delta;
96
97 while (len > 0) {
98 bsize = min(bsize, len);
99
100 if ((error = ffs_balloc1(vp, lblkno(fs, off), bsize, ap->a_cred,
101 ap->a_flags, NULL))) {
102 return error;
103 }
104
105 /*
106 * increase file size now, VOP_BALLOC() requires that
107 * EOF be up-to-date before each call.
108 */
109
110 if (ip->i_ffs_size < off + bsize) {
111 ip->i_ffs_size = off + bsize;
112 if (vp->v_uvm.u_size < ip->i_ffs_size) {
113 uvm_vnp_setsize(vp, ip->i_ffs_size);
114 }
115 }
116
117 off += bsize;
118 len -= bsize;
119 }
120 return 0;
121 }
122
123 int
124 ffs_balloc1(vp, lbn, size, cred, flags, bpp)
125 struct vnode *vp;
126 ufs_daddr_t lbn;
127 int size;
128 struct ucred *cred;
129 int flags;
130 struct buf **bpp;
131 {
132 struct inode *ip = VTOI(vp);
133 struct fs *fs = ip->i_fs;
134 ufs_daddr_t nb;
135 struct buf *bp, *nbp;
136 struct indir indirs[NIADDR + 2];
137 ufs_daddr_t newb, *bap, pref;
138 int deallocated, osize, nsize, num, i, error;
139 ufs_daddr_t *allocib, *blkp, *allocblk, allociblk[NIADDR + 1];
140 UVMHIST_FUNC("ffs_balloc"); UVMHIST_CALLED(ubchist);
141
142 UVMHIST_LOG(ubchist, "vp %p lbn 0x%x size 0x%x", vp, lbn, size,0);
143 if (bpp != NULL) {
144 *bpp = NULL;
145 }
146 if (lbn < 0)
147 return (EFBIG);
148 fs = ip->i_fs;
149
150 /*
151 * If the file currently ends with a fragment and
152 * the block we're allocating now is after the current EOF,
153 * this fragment has to be extended to be a full block.
154 */
155
156 nb = lblkno(fs, ip->i_ffs_size);
157 if (nb < NDADDR && nb < lbn) {
158 osize = blksize(fs, ip, nb);
159 if (osize < fs->fs_bsize && osize > 0) {
160 error = ffs_realloccg(ip, nb,
161 ffs_blkpref(ip, nb, (int)nb, &ip->i_ffs_db[0]),
162 osize, (int)fs->fs_bsize, cred, bpp, &newb);
163 if (error)
164 return (error);
165 ip->i_ffs_size = lblktosize(fs, nb + 1);
166 uvm_vnp_setsize(vp, ip->i_ffs_size);
167 ip->i_ffs_db[nb] = ufs_rw32(newb,
168 UFS_MPNEEDSWAP(vp->v_mount));
169 ip->i_flag |= IN_CHANGE | IN_UPDATE;
170
171 if (bpp) {
172 if (flags & B_SYNC)
173 bwrite(*bpp);
174 else
175 bawrite(*bpp);
176 }
177 else {
178 /*
179 * XXX the data in the frag might be
180 * moving to a new disk location.
181 * we need to flush pages to the
182 * new disk locations.
183 * XXX we could do this in realloccg
184 * except for the sync flag.
185 */
186 (vp->v_uvm.u_obj.pgops->pgo_flush)
187 (&vp->v_uvm.u_obj, lblktosize(fs, nb),
188 lblktosize(fs, nb + 1),
189 flags & B_SYNC ? PGO_SYNCIO : 0);
190 }
191 }
192 }
193 /*
194 * The first NDADDR blocks are direct blocks
195 */
196 if (lbn < NDADDR) {
197
198 nb = ufs_rw32(ip->i_ffs_db[lbn], UFS_MPNEEDSWAP(vp->v_mount));
199 if (nb != 0 && ip->i_ffs_size >= lblktosize(fs, lbn + 1)) {
200
201 /*
202 * the block is an already-allocated direct block
203 * and the file already extends past this block,
204 * thus this must be a whole block.
205 * just read the block (if requested).
206 */
207
208 if (bpp != NULL) {
209 error = bread(vp, lbn, fs->fs_bsize, NOCRED,
210 &bp);
211 if (error) {
212 brelse(bp);
213 return (error);
214 }
215 *bpp = bp;
216 }
217 return (0);
218 }
219 if (nb != 0) {
220
221 /*
222 * Consider need to reallocate a fragment.
223 */
224
225 osize = fragroundup(fs, blkoff(fs, ip->i_ffs_size));
226 nsize = fragroundup(fs, size);
227 if (nsize <= osize) {
228
229 /*
230 * the existing block is already
231 * at least as big as we want.
232 * just read the block (if requested).
233 */
234
235 if (bpp != NULL) {
236 error = bread(vp, lbn, osize, NOCRED,
237 &bp);
238 if (error) {
239 brelse(bp);
240 return (error);
241 }
242 *bpp = bp;
243 }
244 return 0;
245 } else {
246
247 /*
248 * the existing block is smaller than we want,
249 * grow it.
250 */
251
252 error = ffs_realloccg(ip, lbn,
253 ffs_blkpref(ip, lbn, (int)lbn,
254 &ip->i_ffs_db[0]), osize, nsize, cred,
255 bpp, &newb);
256 if (error)
257 return (error);
258 }
259 } else {
260
261 /*
262 * the block was not previously allocated,
263 * allocate a new block or fragment.
264 */
265
266 if (ip->i_ffs_size < lblktosize(fs, lbn + 1))
267 nsize = fragroundup(fs, size);
268 else
269 nsize = fs->fs_bsize;
270 error = ffs_alloc(ip, lbn,
271 ffs_blkpref(ip, lbn, (int)lbn, &ip->i_ffs_db[0]),
272 nsize, cred, &newb);
273 if (error)
274 return (error);
275 if (bpp != NULL) {
276 bp = getblk(vp, lbn, nsize, 0, 0);
277 bp->b_blkno = fsbtodb(fs, newb);
278 if (flags & B_CLRBUF)
279 clrbuf(bp);
280 *bpp = bp;
281 }
282 }
283 ip->i_ffs_db[lbn] = ufs_rw32(newb, UFS_MPNEEDSWAP(vp->v_mount));
284 ip->i_flag |= IN_CHANGE | IN_UPDATE;
285 return (0);
286 }
287
288 /*
289 * Determine the number of levels of indirection.
290 */
291
292 pref = 0;
293 if ((error = ufs_getlbns(vp, lbn, indirs, &num)) != 0)
294 return(error);
295 #ifdef DIAGNOSTIC
296 if (num < 1)
297 panic ("ffs_balloc: ufs_bmaparray returned indirect block\n");
298 #endif
299 /*
300 * Fetch the first indirect block allocating if necessary.
301 */
302 --num;
303 nb = ufs_rw32(ip->i_ffs_ib[indirs[0].in_off],
304 UFS_MPNEEDSWAP(vp->v_mount));
305 allocib = NULL;
306 allocblk = allociblk;
307 if (nb == 0) {
308 pref = ffs_blkpref(ip, lbn, 0, (ufs_daddr_t *)0);
309 error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize,
310 cred, &newb);
311 if (error)
312 return (error);
313 nb = newb;
314 *allocblk++ = nb;
315 bp = getblk(vp, indirs[1].in_lbn, fs->fs_bsize, 0, 0);
316 bp->b_blkno = fsbtodb(fs, nb);
317 clrbuf(bp);
318 /*
319 * Write synchronously so that indirect blocks
320 * never point at garbage.
321 */
322 if ((error = bwrite(bp)) != 0)
323 goto fail;
324 allocib = &ip->i_ffs_ib[indirs[0].in_off];
325 *allocib = ufs_rw32(nb, UFS_MPNEEDSWAP(vp->v_mount));
326 ip->i_flag |= IN_CHANGE | IN_UPDATE;
327 }
328 /*
329 * Fetch through the indirect blocks, allocating as necessary.
330 */
331 for (i = 1;;) {
332 error = bread(vp,
333 indirs[i].in_lbn, (int)fs->fs_bsize, NOCRED, &bp);
334 if (error) {
335 brelse(bp);
336 goto fail;
337 }
338 bap = (ufs_daddr_t *)bp->b_data;
339 nb = ufs_rw32(bap[indirs[i].in_off],
340 UFS_MPNEEDSWAP(vp->v_mount));
341 if (i == num)
342 break;
343 i += 1;
344 if (nb != 0) {
345 brelse(bp);
346 continue;
347 }
348 if (pref == 0)
349 pref = ffs_blkpref(ip, lbn, 0, (ufs_daddr_t *)0);
350 error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, cred,
351 &newb);
352 if (error) {
353 brelse(bp);
354 goto fail;
355 }
356 nb = newb;
357 *allocblk++ = nb;
358 nbp = getblk(vp, indirs[i].in_lbn, fs->fs_bsize, 0, 0);
359 nbp->b_blkno = fsbtodb(fs, nb);
360 clrbuf(nbp);
361 /*
362 * Write synchronously so that indirect blocks
363 * never point at garbage.
364 */
365 if ((error = bwrite(nbp)) != 0) {
366 brelse(bp);
367 goto fail;
368 }
369 bap[indirs[i - 1].in_off] = ufs_rw32(nb,
370 UFS_MPNEEDSWAP(vp->v_mount));
371 /*
372 * If required, write synchronously, otherwise use
373 * delayed write.
374 */
375 if (flags & B_SYNC) {
376 bwrite(bp);
377 } else {
378 bdwrite(bp);
379 }
380 }
381 /*
382 * Get the data block, allocating if necessary.
383 */
384 if (nb == 0) {
385 pref = ffs_blkpref(ip, lbn, indirs[i].in_off, &bap[0]);
386 error = ffs_alloc(ip, lbn, pref, (int)fs->fs_bsize, cred,
387 &newb);
388 if (error) {
389 brelse(bp);
390 goto fail;
391 }
392 nb = newb;
393 *allocblk++ = nb;
394 bap[indirs[i].in_off] = ufs_rw32(nb,
395 UFS_MPNEEDSWAP(vp->v_mount));
396 /*
397 * If required, write synchronously, otherwise use
398 * delayed write.
399 */
400 if (flags & B_SYNC) {
401 bwrite(bp);
402 } else {
403 bdwrite(bp);
404 }
405 if (bpp != NULL) {
406 nbp = getblk(vp, lbn, fs->fs_bsize, 0, 0);
407 nbp->b_blkno = fsbtodb(fs, nb);
408 if (flags & B_CLRBUF)
409 clrbuf(nbp);
410 *bpp = nbp;
411 }
412 return (0);
413 }
414
415 brelse(bp);
416
417 if (bpp != NULL) {
418 if (flags & B_CLRBUF) {
419 error = bread(vp, lbn, (int)fs->fs_bsize, NOCRED, &nbp);
420 if (error) {
421 brelse(nbp);
422 goto fail;
423 }
424 } else {
425 nbp = getblk(vp, lbn, fs->fs_bsize, 0, 0);
426 nbp->b_blkno = fsbtodb(fs, nb);
427 clrbuf(nbp);
428 }
429 *bpp = nbp;
430 }
431 return (0);
432 fail:
433 /*
434 * If we have failed part way through block allocation, we
435 * have to deallocate any indirect blocks that we have allocated.
436 */
437 for (deallocated = 0, blkp = allociblk; blkp < allocblk; blkp++) {
438 ffs_blkfree(ip, *blkp, fs->fs_bsize);
439 deallocated += fs->fs_bsize;
440 }
441 if (allocib != NULL)
442 *allocib = 0;
443 if (deallocated) {
444 #ifdef QUOTA
445 /*
446 * Restore user's disk quota because allocation failed.
447 */
448 (void)chkdq(ip, (long)-btodb(deallocated), cred, FORCE);
449 #endif
450 ip->i_ffs_blocks -= btodb(deallocated);
451 ip->i_flag |= IN_CHANGE | IN_UPDATE;
452 }
453 return (error);
454 }
455